On the Convergence of Tsetlin Machines for the AND and the OR Operators
Keywords: Tsetlin Machine, Convergence, AND operator, OR operator
TL;DR: Main text: 9 pages, Appendices: 19 pages, all combined into a single file for easier reference.
Abstract: The Tsetlin Machine (TM) is an innovative machine learning algorithm rooted in propositional logic, achieving state-of-the-art performance in various pattern recognition tasks. While previous studies analyzed its convergence properties for the 1-bit and XOR operators, this work extends the analysis to the AND and OR operators, completing the study of fundamental digital operations. Our findings demonstrate that the TM almost surely converges to reproduce the AND and OR operators when trained on noise-free data over an infinite time horizon. Notably, the analysis of the OR operator uncovers a distinct property: the ability of the TM to represent two sub-patterns jointly within a single clause, contrasting with its behavior in the XOR case. Furthermore, we investigate the TM’s behavior for AND/OR/XOR operators with noisy training samples, including mislabeled samples and irrelevant inputs. With wrong labels, the TM does not converge to the intended operators but can still learn efficiently. With irrelevant variables, the TM converges to the intended operators almost surely. Together, these analyses provide a comprehensive theoretical foundation for the TM's convergence properties across basic Boolean operators.
Primary Area: other topics in machine learning (i.e., none of the above)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 7748
Loading